1. Field of the Invention
The present invention relates to the configuration of an image acquisition apparatus for measuring the shape of the surface of an object utilizing a two-dimensional array type confocal optical system.
2. Description of the Background
The position (hereinafter referred to as the height) of the object measured in the direction of the optical axis (hereinafter referred to as the z-axis direction) can be measured accurately by utilizing a confocal optical system. Before description of the prior art, the principle of measurement of height by a confocal optical system is explained. A basic configuration of the confocal optical system is shown in FIG. 8. Light emitted from a point light source 81 passes through a half-mirror 82 and is refracted by an objective lens 83 so as to converge onto the object. The light that is reflected by the object reenters the objective lens 83 and is caused to converge by the objective lens 83 and then diverted by the half-mirror 82 toward a pinhole 84 disposed at the same position optically as the point light source 81. The amount of light that passes through the pinhole 84 is detected by a photodetector 85. This is the basic configuration of the confocal optical system. By using this optical system, the height of points on the surface of the object can be measured in the following manner. If the surface of the object that reflects the light is located at a position conjugate to the point light source 81, the reflected light focuses at the position of the pinhole 84. Therefore, almost all the light diverted by the half-mirror 82 passes through the pinhole 84. As the surface that reflects light for illumination is at a greater distance away from the position conjugate to the point light source, the amount of light that passes through the pinhole 84 sharply decreases. This makes it possible to calculate the height of the surface by moving the objective lens 83 and determining the position where the output of the photodetector 85 becomes a maximum. This is the principle of measurement of height by a confocal optical system.
A confocal optical system of the above basic configuration can measure only one point on the surface of the object. For three-dimensional shape measurement, two-dimensional measurement of an appropriate horizontal area is required. To obtain two-dimensional confocal data (hereinafter referred to as confocal images), a confocal image acquisition apparatus utilizing the above confocal optical system must have a scanning means or a plurality of confocal optical systems disposed in parallel. A typical image acquisition apparatus of the latter is a two-dimensional array type image acquisition apparatus. A two-dimensional array type image acquisition system has a feature that can capture a whole two-dimensional image at one time thereby making very fast measurement possible. A two-dimensional array type image acquisition apparatus is disclosed in the specification of Japanese patent application No. 94682/1996. This apparatus is described below with reference to FIG. 7 as a representative example of a two-dimensional array type image acquisition apparatus.
Light for illumination that is emitted from a light source 1 passes through an illuminating pinhole 2 and is refracted by an illuminating lens 3 into parallel rays of light, then enters a beam splitter 4. The light that passes through the beam splitter 4 falls upon a microlens array 5 and is caused to converge by individual lenses of the microlens array 5 to form a very small spot at their focal points. A pinhole array 6 is placed at the position of the focal plane of the microlens array 5. The pinhole array 6 is aligned with the microlens array 5 so that each pinhole of the pinhole array 6 is centered on the optical axis of the corresponding lens of the microlens array 5 at the focal point. Therefore, almost all the illumination light that passes through the lenses of the microlens array 5 passes through the pinholes of the pinhole array 6. The pinholes of the pinhole array 6 are equivalent to point light sources arrayed in parallel. The illumination light that passes through the pinhole array 6 is caused to converge by a bidirectional telecentric objective lens 7 consisting of lenses 7a and 7b and a diaphragm 8 to form an image of the pinhole array 6 (a large number of arrayed small spots) on the surface of an object A. The light that is reflected from each spot on the surface of the object enters the objective lens 7 and is caused to converge by the objective lens 7 toward the corresponding pinhole of the pinhole array 6. The pinholes of the pinhole array 6 also perform the same function as the pinhole in front of the photodetector of the above described basic single-point confocal optical system. If the surface of the object A is on the focal plane of the pinhole array 6, the light reflected from the spots on the surface is focused just at the corresponding pinholes of the pinhole array 6, and the greatest amount of light passes through the pinholes. As the surface of the object A is away from the focal plane of the pinhole array 6, the amount of light that passes through the pinholes sharply decreases. The reflected light that passes through each pinhole of the pinhole array 6 is refracted by the corresponding microlens of the microlens array 5 into a parallel-ray light beam. The light beams are then diverted by the beam splitter 4 and pass through an image re-forming lens 9 to form a reduced image of the microlens array 5 on a two-dimensional arrayed photodetector 10. The two-dimensional arrayed photodetector 10 detects the intensity of individual light beams and outputs an electric signal proportional to the intensity.
By this configuration, it is possible to capture all points of a confocal image simultaneously. Therefore, the shape of the surface of an object can be measured by changing the distance between the objective lens and the object to capture confocal images at different Z positions and to find the position where the intensity of the pixel is greatest for each pixel.
A two-dimensional array type image acquisition apparatus is suited for industrial use because of its high measurement speed enabled by simultaneous focusing of all points to be measured, having no moving parts, and a high-speed shutter camera usable for the photodetector.
However, a two-dimensional array type image acquisition apparatus has a problem of speckles. Since a confocal imaging system is basically a coherent imaging system and illumination light shone on the surface of the object has a high coherency within each spot, the light reflected from each spot interferes to cause speckles in and around the corresponding pinhole of the pinhole array if there are inequalities in the surface within the spot area. Such speckles have a very bad effect on height measurement.
A solution to this problem is disclosed in the specification of Japanese patent application No. 55485/1997. The invention solves the problem by moving illumination spots in a small back and forth scanning motion (vibration) that does not overlap the scanning motion of the adjacent spots during each exposure time of the two-dimensional arrayed photodetector in one complete measurement, utilizing the fact that adjacent spots are seperated by a distance several times the diameter of the spots in a two-dimensional array type confocal image acquisition apparatus. If inequalities in the surface are random, the scanning motion of the spots causes speckles to flicker randomly and averages the intensity of the reflected light passing through the pinholes.
There is another problem in addition to the problem of speckles. Since there is a comparatively large space between adjacent illumination spots, the information about the regions on the surface between spots is not obtained. An image of an object captured by a combination of an ordinary image-forming optical system and a two-dimensional arrayed photodetector such as a CCD camera contains information on almost the whole surface of the object, since there is only a very small space between adjacent pixels. Each pixel of an image represents the total amount of light reflected from the corresponding region of the surface, though averaged over the region. Therefore, even a projection smaller than a pixel (bright spot) can be detected. On the other hand, a two-dimensional array type confocal image acquisition apparatus cannot measure such a projection if it is located between illumination spots.
This problem is described below using a concrete example. A package for a semiconductor chip called a BGA (Ball Grid Array) or a CSP (Chip Scale Package) is used. This package has ball-shaped electrodes arrayed on a plane as shown in FIG. 9. It has suddenly come into wide use in recent years because of the small mounting area required for the number of electrodes. It is important for this package that the tops of all balls are aligned at the same height. Therefore, the inspection apparatus must accurately measure the height of the tops of the balls. Since the balls have a near-mirror surface, only the light reflected from the top part of the balls returns to the objective lens if the balls are illuminated from directly above. The size of the top part that reflects the light returning to the objective lens varies depending on the diameter of the balls and the NA of the objective lens. It decreases to a size smaller than a pixel as the diameter of the balls becomes smaller. An image acquisition apparatus using an ordinary image-forming optical system and a two-dimensional arrayed photodetector such as a CCD camera can measure even the top part of the balls smaller than a pixel, though not accurately. On the other hand, a two-dimensional array type confocal image acquisition apparatus cannot measure this unless an illumination spot is directed onto it.
In brief, a conventional two-dimensional array type confocal image acquisition apparatus has a problem inasmuch as it cannot measure a small object because of blind regions between illumination spots.
In addition to the above problem that a small object cannot be measured, blind regions between illumination spots cause another problem. Subpixel-accuracy edge measurement, commonly possible when performing measurement of the edges of objects in images by image processing, is not possible by image processing of images obtained by a two-dimensional array type confocal image acquisition apparatus because of blind regions between illumination spots, or a low aperture ratio for pixels resulting from blind regions.